The room's floor is not a number, it is a range: 33 dB(A) with the air handler idle and 44 dB(A) while it runs, which puts the room in two different bands depending on when you measure.
An HVAC duty cycle makes a one-minute reading a lie
written 2026-09-02rots yes — figures above moveresponsible Room Tone Register (a named register, not a person)
What was measured
A treated voice room, 27 m³, door closed, building otherwise unoccupied. Meter on a stand at the microphone position, A-weighted, slow, one reading every thirty seconds for twelve minutes. The trace has two states and no intermediate values: 33 dB(A) idle, 44 dB(A) running, with the running period lasting roughly four minutes in every twelve.
The judgement
A sixty-second reading taken during the idle period reports 33 dB(A) and puts the room in the band that permits close-mic'd voice work. A reading taken during the running period reports 44 dB(A) and puts it in the band that permits only loud sources. Both readings are correct. The room is in both bands, and the one that matters is the one that will be true during the take.
| Reading window | Reported | Band it implies | Wrong? |
|---|---|---|---|
| 60 s, idle | 33 dB(A) | Under 30 to 40 | Yes, if the take runs long |
| 60 s, running | 44 dB(A) | 40 to 50 | Yes, if you can switch it off |
| 12 min, equivalent | 40.4 dB(A) | 40 to 50 | No, this is the honest number |
| 12 min, worst state | 44 dB(A) | 40 to 50 | This is the number to plan for |
What follows for room tone
It follows that a single room tone take is not enough in a room with a cycling source. The take will contain whichever state was running. When the editor needs to cover a seam from a take recorded in the other state, the residual will not match and the seam will be audible. The fix is cheap and takes ninety seconds: record a pass with the air handler on and a pass with it off, slate both, and let the editor choose.
This is also why 'the room measured 33' is a sentence worth distrusting. Ask how long the measurement ran, and what the air handler was doing. If the answer is a minute and 'I do not know', the number is a sample, not a measurement.
Source log
- 0129 CFR 1910.95, Occupational noise exposure (eCFR)supports the exposure framework that makes a stated averaging time necessary
- 021910.95 Appendix A: Noise exposure computation (OSHA)supports the computation of an equivalent level from a set of readings
- 03Frequency-Dependent Damping, CCRMA (Stanford)supports the frequency-dependent behaviour that makes a cycling source audible under dialogue
Each line points at one specific document. No line is a home page.
Leave this page Close this page if your room has no cycling source — the single-state case is covered on the noise floor page.